Speech-Recognition Cloud Harvesting for Improving the Navigation of Cyber-Physical Wheelchairs for Disabled Persons

Abstract A cyber-physical system for a speech-controlled wheelchair has been proposed. It is based on cloud-harvesting principles, which result in significant improvement in command error rate (CER). The overall methodology and developed cloud-harvesting algorithm have been presented, discussed, and tested. The combination of IBM Watson and Google Cloud Speech APIs gave significantly better results than the use of solitary speech recognition APIs. The proposed approach shows the potential for and usability of speech-controlled wheelchairs, as well as in similar applications.